PROJECT SUMMARY Hepatocellular carcinoma (HCC) is the second leading cause of cancer-related death worldwide with more than 745,000 fatalities in 2012 alone. The majority of patients with HCC present with unresectable disease at diagnosis and a life expectancy of less than 20 months, with only 15% of these patients surviving more than one year after diagnosis. This dismal prognosis underscores the limited therapeutic options for these patients. HCC is a notoriously chemoresistant malignancy and advances in targeted therapeutics have been unsuccessful in improving survival resulting in what some have called ?a losing battle? in the development of therapies. This deficiency issues, in large part, from limitations of current preclinical models in (i) recapitulating the inter- and intra-tumoral heterogeneity that characterizes HCC and (ii) predicting patient response to therapeutics. While patient-derived tumor models have been demonstrated to more faithfully recapitulate the heterogeneity of human tumors, there has been limited validation of the translational relevance of these models with respect to their fidelity to the intra- and inter-tumoral mutational heterogeneity that characterizes HCC as well as their ability to provide translationally reliable information for the design, testing and/or outcome evaluation of novel or existing therapies. Indeed, the creation of new patient-derived models of HCC requires rigorous validation of the resulting tumors to confirm their fidelity to the cancer of interest and robust credentialing criteria to ascertain their biological relevance and reliability as surrogates of patient response. In preliminary studies we have: 1) demonstrated the ability to generate PDXs and PDX-derived cell lines from percutaneous biopsies of tumors in patients with intermediate stage HCC and 2) developed methodologies to enable the characterization, validation and optimization of these models. The proposed project will build on this prior work to assess the fidelity and predictive potential of patient-derived models of HCC. We hypothesize that patient-derived models of HCC derived from percutaneous biopsies recapitulate the inter- and intra-tumoral heterogeneity of their parent biopsies and that these models are predictive of patient response to therapy. To test this hypothesis the proposed project will pursue three aims: (1) to define the representation of inter- and intra-tumoral clonal heterogeneity of patient-derived models of HCC through targeted sequencing and digital polymerase chain reaction; (2) to determine the predictive potential of patient-derived models of HCC for response to common HCC therapies; and (3) to investigate the role of HCC tumor initiating cells (TICs) in improving the yield and predictive potential of patient-derived models of HCC. Importantly, the achievement of the proposed aims will transform the utility of patient-derived models of HCC for translational research.